Monday, June 23, 2014

A New Look at BITX Carrier Suppression; N6QW IN EMRFD!

Bert, WF7I, was recently struggling to get his BITX 20 going and he asked some good questions about carrier suppression.  I realized that I hadn't really paid much attention to this.  Perhaps as a result of my long experience with DSB, I was happy as long as I was able to null out MOST of the carrier. 

I fired up the scope and took a look at the output from the BITX 2040 on 40 meters.  Here's the test setup:  Coax from the antenna terminal to a 50 ohm resistive load at the Rigol O'scope probe.  Just keying the transmitter (no mic connected),  carrier was at 980 millivolts rms or about 19 milliwatts.  I then connected an AF sig generator into the mic in connector and pumped in some 1000 Hz sine wave. Peak output was 20.7 volts rms, or about 8.6 watts.   That puts the carrier about 27 db down.  I felt I should be doing better. 

I took a look at the shape of my crystal filter and the frequency placement of my carrier oscillator.  I noticed that the carrier oscillator freq was fairly close to the bandpass portion of the crystal filter -- fairly high up the skirt, only about 9 db below the passband level.   I figured that if I just moved that carrier oscillator up around 300 Hz, I would get around 10 db of additional carrier suppression. 

Sure enough,  with the carrier moved a mere 300 Hz further away from the passband,  the residual carrier dropped to 346 millivolts rms, or about 2.4 milliwatts.    Now peak output was 20.9 volts rms, or 8.7 watts.  36 db of carrier suppression. 

I guess I could do better if I moved the carrier up another little bit, but I like the sound of it now.  I may have been able to better if I'd fiddled with the balanced modulator diodes a bit more.  But what do you guys think?  Should I worry about 2 milliwatts of residual carrier?  Heck I once ran a CW rig (W1VD' VXO 6 watter) that kept the oscillator running on key up, producing about 15 mw of "backwave." No damage was done, few noticed, no one complained.

Oh yea, is this the way to measure carrier suppression? 

----

While doing all this, I pulled out my trusty copy of EMRFD.  The index led me to the balanced modulator section on page 6.56.  There I spotted a familiar call:  W6JFR!!!  That's Pete Juliano, N6QW!  Pete is credited with a mod to the SBL-1 mixer that adds a balance control pot to the device.   Wow, actually being IN EMRFD fully confirms Pete's homebrew guru status.


Our book: "SolderSmoke -- Global Adventures in Wireless Electronics" http://soldersmoke.com/book.htm Our coffee mugs, T-Shirts, bumper stickers: http://www.cafepress.com/SolderSmoke Our Book Store: http://astore.amazon.com/contracross-20

4 comments:

  1. Hi again Bill.

    Leonard, KC0WOX, was helping me over on the Bitx Yahoo group. I'll re-post the link he posted there:
    http://golddredgervideo.com/kc0wox/bitxver3/version3balmod.wmv

    He walks through his testing of the bitx "version 3" balanced modulator(a different kit than the one I built) and with it hooked up to a spectrum analyzer he reports 50-60 dB suppression (20 W out on the SSB signal), which seems quite good to me and better than what I obtained. You can keep sliding the BFO down in freq to improve this but after awhile the speech gets pretty ugly, sounds like a serious speech processor to me.

    I really appreciated that Leonard took the time to post such detailed vids of his testing, it helped me verify how mine was working in comparison with someone else.

    Another trick which I found fun to play with and easy to do is to find diodes with closer matched characteristics for the balanced modulator. At first I thought the variable cap and resistor ought to nullify slight differences in turn-on voltages with diodes and that didn't matter. But in my own testing I found a slight improvement in the carrier null I was able to achieve by just picking different diodes.

    Unfortunately I didn't do very systematic testing of this but I'd say maybe another 5 or even 10 dB might be possible, if such things matter (and I'd appreciate others to weigh in on "how much carrier should we suppress"). For my kit board, having less carrier suppression seemed to instigate oscillation problems in the PA taking place at lower output powers (which I guess makes sense). But maybe with a different layout and better isolation, that doesn't really matter.

    73,
    Bert WF7I

    ReplyDelete
  2. whoops, the "diode trick" was posted/mentioned by Arv and Leonard over on the Yahoo group -- not my idea (just trying to give proper credit!).

    -- Bert WF7I

    ReplyDelete
  3. I think one problem in the old days was heat, no matter how you balanced things, they'd come unbalanced when the tubes heated up, or the components aged.

    It wasn't uncommon to see a resistive balance, and a trimmer capacitor in the old circuits. Now, there seems to be less fussing, not sure if the circuits are better, or the components better matched, or what.

    There was one commercial phasing rig that had a notch filter at the carrier frequency, maybe a reflection of the balanced modulators of the day, but also no filter to knock down the carrier.

    With everyone buying lots of crystals on the same frequency in order to make ladder filters, maybe one of the surplus crystals would work as a notch filter at the output of the balanced modulator, maybe not much needed beyond the crustal, it would just be there to add that tiny bit more of carrier suppression.

    Reminds me of a 1974
    in Ham Radio about building an SSB IF strip with TTL logic ICs, an intriguing idea but I suspect nobody went further with it. But they just used a gate as a mixer, so no ability to balance out the carrier. Instead,it relied on a crystal at the output to notch out the carrier.

    Michael VE2BVW

    ReplyDelete
  4. HI,

    My target number has been not less than 40db and more like 50,

    Layout is important as leakage past the filter or more correctly around the filter is problematic. The BITX layouts I've seen put the Carrier o/c near the IF enough that it can blow by the filter.

    The balanced mod should hit 30db alone and some will do as high as 60 (mc1496) but practical is in the 45 to 50 realm. The filter should add about 10db to that, This means that the balanced mod should have both a resistive balance and capacitive balance. the key is balanced not adjusted to minimum.

    As to method many work. Since I build and often adjust before completion (milliwatt maybe) that is like a first light adjustment to insure the balanced mod is functional and balanced. I generally do it with a diode probe. adjust for balance and insure its a minima. To Do that I often offset the carrier os to the middle of the filter bandwidth or even jumper out the filter so that I have gain before the probe to see the really weak signal. I can then reset the carrier to the right point for the filter.

    Its tricks. Much of it learned while building my first examples over 40 years ago.

    Allison/KB1GMX

    ReplyDelete